Thai Journal of Operations Research : TJOR

Analysis of Probability and Risk in Baccarat Through Computer Simulation

2025-11-25T15:56:29+07:00

This study aims to analyze the probability and assess the risk of the Baccarat game using simulations in R. The research evaluates the likelihood of winning across different betting options and calculates the expected value for each scenario. By simulating 10,000 rounds for each betting configuration, the results indicate that the probability of winning is consistently lower than the probability of losing in all cases. Furthermore, all betting strategies yield a negative expected value, placing players at a disadvantage in the long term. Betting equally on Player and Banker results in the most negative expected value. Although increasing bets on Tie slightly reduces the negative expectations, it remains an unfavorable strategy in the long run. Additionally, the findings confirm that playing Baccarat as a form of gambling cannot generate sustainable profits, as the probability of going bankrupt is 100%. This is because the likelihood of losing a bet always exceeds the chance of winning. The risk assessment further validates that the rate of capital loss increases exponentially as the number of betting rounds increases.

Improvement of the Metal Box and Printed Circuit Board Assembly Process to Reduce Resin Bleed Using Design of Experiments

2026-04-03T10:59:58+07:00

In the metal box and printed circuit board assembly process of the case study company, a contract manufacturer of optical transceiver devices, resin bleed exceeding 5.4 millimeters was identified as a critical issue. This problem led to scrap in the gold wire bonding process. The average resin bleed length before improvement was 5.2633 millimeters. This research aimed to reduce the resin bleed length and improve the bonding quality between metal box and printed circuit board. The average shear force before improvement was 11.3946 gram-force. The study investigated the factors affecting resin bleed length and shear force to determine the optimum factors that minimize resin bleed length and maximize shear force. A central composite design was applied. Four factors were studied: epoxy weight, waiting time after assembly at room temperature, surface cleanliness, and epoxy type. The experimental results were analyzed using response surface methodology. The experimental results indicate that the factors affecting resin bleed length are the interaction between waiting time after assembly at room temperature and epoxy type, and the interaction between surface cleanliness and epoxy type. For shear strength, the significant factors include the quadratic term of epoxy weight, the interaction between epoxy weight and surface cleanliness, the interaction between epoxy weight and epoxy type, and the interaction between surface cleanliness and epoxy type. The optimal conditions were: settings were an epoxy weight of 6.91 milligrams, the waiting time after assembly of 41.72 minutes, plasma surface cleanliness, and epoxy type 2. Confirmation experiments with 30 samples demonstrated that, at a 95 percent confidence level, the mean resin bleed length was less than 1.3461 millimeters, representing a reduction of at least 30 percent compared to the pre-improvement condition. Furthermore, the mean shear force was greater than 27.7776 gram-force, which is significantly higher than the pre-improvement condition.

Multi-warehouse Relief Aid Pre-position Planning for Thai Red Cross

2026-04-05T11:18:08+07:00

Relief supply pre-positioning is essential for effective disaster management, particularly when demand is uncertain and varies across regions. This study proposes a data-driven approach to analyze and forecast the demand for relief aid packages of the Thai Red Cross to support resource planning at the cluster level of Red Cross stations and improve the efficiency of storage and distribution. Monthly demand data from Red Cross stations and the Disaster Relief Division during 2011–2023 were analyzed. Due to missing records in certain periods and changes in station service areas, the dataset was restructured by reallocating historical demand to the original coverage areas to maintain data continuity prior to the analysis. Statistical analysis using the Kolmogorov–Smirnov test indicated that the demand exhibited a right-skewed pattern and was best represented by a lognormal distribution. Based on this distribution, demand levels corresponding to an 80% confidence level were estimated and compared between decentralized station-level storage and cluster-based management, showing that demand aggregation at the cluster level reduced the total safety stock required by the system. For forecasting, several Machine Learning techniques were applied with time, location, and rainfall variables serving as model inputs. where rainfall was incorporated as weighting factor during model training to reflect disaster risk conditions. Experimental results indicate that cluster-based inventory planning significantly reduces overall stock requirements compared with independent station storage, while the selected forecasting model provides reliable demand estimates. Evaluation using the cumulative distribution function (CDF) shows that the proposed approach can support demand coverage of approximately 80–90% at the cluster level and about 60–80% at the station level.

An Explainable Machine Learning Framework for Analyzing Road Accident Severity during Festival Periods

2026-04-05T15:32:33+07:00

Severe road traffic accidents are a critical issue causing immense loss of life and property, particularly during the Songkran festival, which is characterized by heavy traffic and unique environmental factors. This study focuses on analyzing accident severity using data from the Ministry of Transport (2021–2024) by proposing an Explainable Machine Learning (XAI) framework to enhance predictive transparency and practical application. To address data imbalance, the SMOTE technique was employed, and the performance of Random Forest, Support Vector Machine (SVM), and Logistic Regression models was compared. Empirical results showed that the Random Forest model achieved the best overall performance (Accuracy = 0.711 ± 0.006, AUC = 0.706 ± 0.025), while the SVM demonstrated superior capability in identifying high-severity cases (Sensitivity = 0.701 ± 0.033). Furthermore, SHAP (SHapley Additive exPlanations) analysis revealed that motorcycle involvement, the number of vehicles, and self-caused collisions are critical risk factors. These insights provide a data-driven foundation for relevant agencies to develop proactive safety measures and mitigate festival-related accidents sustainably.

Optimization of Employee Scheduling via Integer Linear Programming with a Standby System: A Case Study of ABC Restaurant

2026-04-03T09:57:22+07:00

The restaurant business has been continuously expanding in response to rapidly changing consumer behavior. Consequently, restaurant operations must consider several key factors, including food quality, cost control, and human resource management. In particular, employee management is a critical factor that directly affects service quality and operating costs. This study develops a scheduling model for part-time employees at ABC Restaurant by applying Integer Linear Programming (ILP) techniques to improve workforce allocation efficiency, reduce labor costs, and achieve a balanced distribution of work shifts among employees. The data used in this study were collected from ABC Restaurant over a period of 10 months, divided into two phases: January to May and August to December 2024. An analysis of the existing scheduling system, which consisted of two work shifts per day, revealed that some employees received no weekly assignments, thereby causing workload imbalances. Consequently, an improved scheduling model was developed by increasing the number of daily shifts to four, incorporating employee availability constraints for each shift, introducing a backup staff system, and adjusting the wage rate from 40 baht per hour to 80 baht per hour. The results indicate that the improved model reduced labor costs from 9,600 baht per week (38,400 baht per month) to 8,640 baht per week (34,560 baht per month). At the same time, the model successfully met staffing requirements for regular employees and enhanced managerial flexibility through the use of reserve staff. This research demonstrates that the proposed model effectively improves the efficiency and flexibility of human resource management and can be appropriately applied to other service-oriented businesses.

Optimization of 5-HMF Synthesis from Sucrose via Organic Acid Adsorption Using Resin through Central Composite Design Coupled with Response Surface Methodology

2026-04-20T10:24:12+07:00

This research aimed to enhance the removal efficiency of levulinic acid, a by-product formed during the synthesis of 5-hydroxymethylfurfural (5-HMF) from sucrose, as its presence negatively affects the purity and quality of 5-HMF. Adsorption using resin was applied together with statistical experimental design to develop a quadratic model and determine the optimal conditions for levulinic acid removal. A Central Composite Design (CCD) based on Response Surface Methodology (RSM) was used, considering four factors: stirring time, resin dosage, initial 5-HMF concentration, and levulinic acid concentration. The percentage removal of levulinic acid was set as the response variable. A total of 30 experimental runs were conducted and analyzed using Minitab v.22.4.0. The results showed that the quadratic model significantly described the relationship between the factors and the response (p < 0.05) with a coefficient of determination (R²) of 70.20%. The quadratic terms of stirring time and resin dosage significantly influenced adsorption efficiency. The optimal conditions were 150 minutes of stirring time, 7.00 g of resin, 0.4001 M initial 5-HMF concentration, and 0.0102 M levulinic acid concentration, resulting in a predicted removal efficiency of 67.70% with a 95% confidence interval of 35.1–100. These results indicate that combining adsorption with statistical experimental design can improve the purification of 5-HMF and support sustainable bio-based chemical production.

Development and Comparative Analysis of Demand Forecasting Models

2026-04-05T11:35:42+07:00

The petrochemical industry plays a vital role in the economy as a key supplier of raw materials for various industries, particularly plastic resin pellets used in manufacturing. However, economic fluctuations and the complexity of demand and supply make plastic resin pellets demand forecasting highly challenging. This study aims to develop and compare forecasting models for plastic resin pellets demand using a case study of a petrochemical manufacturer in Thailand. The product selected for this study was chosen based on the highest total sales during the study period. Data characteristics were examined using the Mann-Kendall Trend Test and the Friedman Test, while variables were selected using Mutual Information (MI). Seven forecasting models were developed and compared, including ARIMA, ARIMAX, XGBoost, and hybrid models comprising ARIMA-XGBoost, ARIMAX-XGBoost, XGBoost-ARIMA, and XGBoost-ARIMAX. Model performance was evaluated using the Mean Absolute Percentage Error (MAPE). The results indicate that, in this case study, hybrid models employing Machine Learning as the primary model tend to provide better forecasting performance than most traditional models. In particular, the XGBoost-ARIMAX model achieved the highest forecasting accuracy among all models considered.

Comparison of Forecasting Methods for Gold Prices under the Russia-Ukraine War

2026-04-05T11:01:38+07:00

In the context of the Russia-Ukraine war that erupted in February 2022, global gold prices have shown an upward trend, driven by gold accumulation policies adopted by certain countries to ensure economic security and increased investor demand for safe-haven assets. This study aims to evaluate the forecasting performance of daily gold prices (XAU/USD) by comparing three forecasting approaches, namely Holt's Exponential Smoothing, the ARIMA method following the Box-Jenkins framework, and the XGBoost machine learning algorithm applied with a Direct multi-step-ahead forecasting strategy to construct gold price forecasting models. The study utilizes 773 daily gold price observations, with the first 80% (618 days) used for model development and training, and the remaining 20% (155 days) reserved for performance evaluation and comparison. The results indicate that XGBoost achieved the highest performance across all three evaluation metrics, with a MAD of 132.55, RMSE of 156.49, and MAPE of 4.96%. The findings suggest that while time series methods are capable of capturing overall trend patterns, they are limited in detecting short-term volatility due to inherent model constraints. In contrast, XGBoost combined with the Direct multi-step-ahead forecasting strategy demonstrates superior ability to handle nonlinear relationships within the data, making it more suitable for forecasting the prices of highly volatile assets during geopolitical crises.